Business School and the KICPA Grind, Seen Again in the AI Era

Business school and the KICPA experience, seen again in the AI era.

Lately, using AI and earning small sums of money, and as the era moves on, there is something I keep feeling.

In the process of using AI to sell, to win development contracts, and to design how my work is structured, unexpected past experiences are exerting a bigger influence than I would have guessed.

In particular, the years studying business administration at university from 2020 to 2026, and the year-plus of immersion in KICPA study in 2024, have become a major foundation for how I deploy AI today.

I properly started AI coding in March 2025.

From the outside, it may look like AI coding experience produced the current results.

But in truth, before that, there was a period spent trying to understand what a company actually is.

I studied financial statements,

studied cost structures,

tried to understand capital structures,

and worked to learn business management and tax structures.

In the end, the conclusion I was able to reach was this:

how does a company move?

A company is not simply an organization that makes and sells products.

It is a single system connecting many stakeholders.

A company secures resources through equity and debt.

With those resources it runs its operations, (the factory)

makes products and services, (production)

and generates revenue through sales activity. (sales)

Then it manages its cost structure to leave operating profit,

and creates cash flow to reinvest in growth.

This structure does not change with time.

Even if you forget everything else, something remains:

the perspective that a company takes assets — debt and equity — and creates cash flow and operating profit.

Accounting standards and detailed calculation methods can blur with time.

But the structural lens for looking at a company remains.

And now, applying AI to real work, that lens is in constant use.

I currently take on and run AI-based development contracts.

Orders keep coming in, so rather than accepting everything, I choose what to take.

The way I work differs somewhat from conventional development.

First, I design the whole.

What structure to build it in,

what features are needed,

what process will produce the deliverable — I lay it all out.

Then I use AI on top of that design.

Every feature the client requested is reflected as a baseline,

and the work proceeds toward a higher level of completeness than the conventional approach.

The work itself feels easy and interesting rather than hard.

Analyzing the problem,

designing the structure,

and producing the result with this new tool called AI — the process is fun.

My current working environment is no special office, either.

I work out of a study café.

It is a structure with no need to maintain an office.

The business is registered as a sole proprietorship at a personal address,

and the computer my mother bought me in university — three to four million won — is the entire foundation of the business.

I set up the design, hand the work to the AI,

and the workload runs at a pace where I can actually go out for a walk and come back to check the results.

Most people look at AI and see only the output.

But what actually matters is the process.

“I made it with AI.”

That one sentence explains nothing.

When someone asks how it was made, there is a great deal of process to explain.

Which tool was chosen in which situation,

in what order the work proceeded,

by what logic things were connected,

which parts a human judged.

Countless choices go in.

None of this comes immediately from reading a book or a manual.

It is tacit knowledge that accumulates through direct experience.

Which tool to use in which situation,

how to approach a problem,

the judgment of what counts as a good result.

This becomes an individual’s moat.

In the AI era, access to the tools themselves keeps widening.

But how to combine the tools and where to apply them is the domain of experience.

Right now, for both contracting and the development process,

I am building automation as well.

The goal is a structure that runs without my direct involvement in every step.

With AI, the scope of work one person can handle changes from what it was.

I am studying a form in which one person can operate the workload structure of a mid-sized firm — and even larger enterprises.

The key is not a structure with more people,

but building a structure where the system does the work.

In business there are many strategies.

The Hyundai-style approach is unconditional advance.

Open the market fast,

move first,

and expand.

The Daewoo-style approach follows once a leader moves.

Confirm the market first,

judge the potential, then enter.

The Samsung-style approach moves after stacking up technology and preparation.

Tap the stone bridge before crossing it.

All three approaches have their place in business.

What matters is carrying both: earning money now and preparing for the future.

Time spent generating revenue with AI now is necessary.

At the same time, research time preparing for the next era is necessary too.

Even the walks are not mere rest.

What is coming next,

what lies beyond it — that is the time to think about these things.

In business, it matters not only to see what is visible now,

but to see the next stage and what lies beyond it.

Another place where business school and KICPA study help is in how I look at corporations.

A sole proprietorship versus a corporation is not merely a question of tax differences.

The owner’s disposition,

the way funds are managed,

the direction of growth — these should decide the choice.

A corporation is a system.

It is easiest to understand the corporate entity as a kind of living organism.

Inside it live many stakeholders.

The CEO ponders how to use the resources inside the company to grow it.

Executives care about surviving long within the organization and being rewarded.

Shareholders want dividends and rising enterprise value.

Creditors aim to recover what they lent, safely.

Customers want good products and services.

The government wants companies to keep the law and contribute to economic activity.

Because each goal differs, the agency problem arises.

The money in the company is not the CEO’s personal money.

It is the corporation’s resource, and many stakeholders are connected to it.

That is why a company needs rules and governance structures.

The AI industry today is not a simple technology race.

Big-tech capital,

the stock market,

state capital,

and the interests of governments and corporations are all connected in it.

AI companies compete while securing enormous capital and talent.

AI is hard to see as merely one technology.

It is a means of making money,

a factory that raises productivity,

and a system that plays the role of a company’s new employee.

AI is a technology made by people.

It is therefore not a fully objective being.

The developers’ judgments,

the company’s direction,

the interests of society and state all enter into it.

AI is, in the end, connected to the domain of language and thought.

That is why for Korea, too, it matters to use AI and to secure indigenous technological capability at the same time.

Looking back: the business study from 2020 to 2026,

the year-plus of KICPA preparation in 2024,

and the AI coding begun in March 2025 each looked like separate experiences.

But seen from now, they connect into a single stream.

Business school was the process of understanding corporate structure.

KICPA study was the process of understanding numbers and financial structure.

AI coding is the process of wielding a new tool of production.

What I do now is the combined form of these three.

Understand how a company moves,

find the problem,

and place the new tool called AI in exactly the right position.

Technology keeps changing.

But the basic structure of business does not.

In the end, what matters is not how many new technologies you know,

but the judgment to decide which structure to place a technology in, and what value it can create there.